1. Introduction
Mucosal melanoma (MM) is a rare disease epidemiologically and molecularly distinct from cutaneous melanoma (CM) developing from melanocytes located in mucosal membranes. Overall, MM represents about 1 to 1.5% of all melanoma cases and 0.03% of all diagnosed cancers [
1,
2,
3]. MM incidence has been reported globally stable over the last 20 years [
4,
5]. MM is diagnosed twice more often in Caucasians than in populations with darker skin colors, including African Americans [
6,
7], but much rarer than in the Asian population. The risk of developing MM increases with age. The majority of patients diagnosed are 60 years of age and older. The median age at diagnosis is 70 years, except for MMs arising in the mouth that affect younger patients more frequently [
6,
8]. Anatomically, it is mostly diagnosed on the mucous membranes of the head and neck (31% to 55%), anus and rectum (17% to 24%) or the vulva and vagina (18% to 40%) and less frequently in the colon, throat, larynx, lungs, urinary tract, cervix, esophagus or gallbladder [
3,
8,
9]. The incidence of MM is over 80% higher in women than in men due to the relatively high number of genital melanomas diagnosed in women [
3]. Risk factors for developing MM are currently unknown since ultraviolet (UV) radiation and viral etiology—cytomegalovirus, Eppstein-Barr virus, human papillomavirus, human herpesvirus—have also been excluded. In MM, no environmental exposures nor carcinogenic viruses have been found pathogenic [
10,
11,
12,
13,
14]. Unlike cutaneous melanoma, which is characterized by a UV signature, MM harbors distinct molecular features, including a lower incidence of v-Raf murine sarcoma viral oncogene homolog B
(BRAF) oncogene mutations but a higher incidence of tyrosine-protein kinase KIT (CD117) oncogene mutations, suggesting different genetic etiologies. In general, MM harbors fewer nucleotide substitutions per cell than CM, but more gene amplifications and structural variants than CM; the cause of this chromosomal instability has not yet been clarified [
3].
Mucosal melanomas are generally diagnosed in an advanced stage, though they are more aggressive and carry a worse prognosis regardless of the stage at diagnosis. The five-year overall survival (OS) rate for MM is only 25% regardless of stage [
15]. Poor treatment results and shorter survival may be associated with a lack of early symptoms or signs, resulting in advanced disease at the time of diagnosis. Insidious anatomical localization, and often amelanotic presentation, result in difficult visual detection and challenging resections with wide, or even negative, margins impossible to achieve. Moreover, rich lymphatic drainage from the mucosal surfaces promote metastases [
3,
8,
15]. MM metastases most often develop in the lungs (54%), liver (35%) and bone (25%) [
16].
Due to its rare occurrence, undefined etiopathogenesis and unpredictable clinical course, there are no specific recommendations of MM treatment. Both the European Society for Medical Oncology (ESMO) and National Comprehensive Cancer Network (NCCN) guidelines focus on the important role of surgery and radiotherapy (RT) in this group of patients [
2,
17]. The preferred therapeutic strategy in MM is still surgical excision. Perioperative RT improves local control but does not improve overall survival (OS), probably because of the high rate of distant relapse [
18,
19,
20]. RT with definitive intent can also provide satisfactory local control. Definitive RT should be considered in patients who are not candidates for extensive surgery or in cases where adequate resection margin cannot be achieved [
21]. Proton therapy and heavy ions deserve special attention [
22]. In a retrospective analysis of a cohort of patients with sinonasal MM, the authors observed a 62% five-year local control rate for proton therapy [
23]. The understanding of multimodal treatment in MM comes mostly from CM data extrapolation. Phase III clinical trials focused only on MM patients are lacking. Phase III trials with all melanoma patients, regardless of subtype, do not provide satisfactory data on MM due to the low number of cases enrolled. The real-world data on the efficacy of antiprogram med cell death 1 (PD-1) therapies in patients with MM are scarce. Although a growing number of studies suggest the significant benefit of RT as a boost for immunotherapy (ITH) in CM, such observations in MM have not been reported. Therefore, this study aims to assess the efficacy of systemic therapy with the emphasis on ITH in the MM patient population treated outside of clinical trials, and to define the efficacy of the immunotherapy-radiotherapy combination in MM.
3. Discussion
Until now, limited evidence has supported the efficacy of anti-PD-1 ITH in MM. In this population of patients, in individuals with the KIT mutation, imatinib was reported to result in significant clinical benefits, while for MM patients whose tumors harbor
BRAF mutations, treatment with a combination of BRAF and MEK inhibitors was shown to be effective [
8,
24]. The efficacy of anti-PD1 ITH remains unclear in MM patients since a low number of patients have been enrolled in clinical trials. Conflicting data on MM patients’ responses to nivolumab, pembrolizumab and ipilimumab have been published. Some reports suggest that anti-PD ITH is significantly less effective in MM than in CM patients, while other report durable anti-tumor effects [
25,
26,
27].
We analyzed anti-PD-1 ITH efficacy in routine clinical practice outside of clinical trials in a nonpreselected population of subsequent patients with MM, and report ORR and PFS similar to patients with CM, also treated outside of clinical trials. In a retrospective analysis covering 25 dermatology departments in France, 75 MM patients were treated with first-line nivolumab or pembrolizumab and, out of these, fifteen achieved OR, which corresponds to an ORR of 20% (95% CI: 11.6–30.8) [
28], which is lower than the 25% reported by us. In a Japanese study of 24 MM cases, the ORR was 20.8%, and for 17 cases with visceral metastases—17.6%. For the 17 cases, OS and PFS periods were 422 days and 226 days [
29]. In another small Japanese trial, 17 MM patients were treated with nivolumab, and ORR was reported at 23.5%. One patient achieved a CR, three—PR, and five SD as their best response. The median PFS was 1.4 months (95% CI: 1.2–2.8) [
30], which is again inferior to the results reported by us. High OR was reported for the first-line pembrolizumab—35% accompanied with five-months of median PFS (
n = 20) [
31].
We previously described the results of anti-PD1 therapy as well as BRAF/MEK inhibitors treatment in the general melanoma population in Poland [
32]. While comparing data between the general population of patients receiving anti-PD1 ITH in the first line and an analogous subpopulation of mucosal melanoma patients, no differences were seen in ORR (objective response rate), PFS and OS. The ORR was equal to 28% in the mucosal melanoma subgroup, and 30% in the general population (
p = 0.85 by Fisher’s exact test). The median PFS and OS for mucosal melanoma was 6.0 (CI: 3.8–10.8) and 15.8 (CI: 11.5–NA) months, while for the general population—6.9 (CI: 5.3–9.0) and 20.5 (CI: 15.3–NA), respectively. The HR for mucosal melanoma was 0.97 (CI: 0.68–1.38,
p = 0.86) for PFS and 0.91 (CI: 0.60–1.38,
p = 0.65) for OS (see
Supplement B). Therefore, our results confirm the efficacy of immunotherapy, especially anti-PD1 antibodies, in MM treatment [
31].
Due to strict inclusion criteria in clinical trials, covering mostly patients without brain metastases and patients with good performance status, corresponding to low tumor burden, we expected longer PFS and OS for patients treated in the trials than for patients treated in our study. In a meta-analysis of major ITH clinical trials by Shoushtari, A.N. et al., 35 MM patients were identified. This analysis covered patients treated within NCT02083484 (MK-3475), NCT01295827 (KEYNOTE-001), NCT01927419 (CheckMate 069), NCT01024231 (CA209-004), and NCT01721746 (CheckMate 037) trials. Therefore, the majority of patients were treated with pembrolizumab or nivolumab, not in the first line but after previous therapy. In these trials the majority of patients received ipilimumab before anti-PD-1 ITH. These authors reported that for MM treated with anti-PD-1 ORR was 23% (95% CI: 10–40%) with PD as the best response for 57% of patients (95% CI: 39–74%), which represents ORR similar as that reported by us (25% for anti-PD1 in the first line). In the same meta-analysis, MM patients treated with nivolumab/pembrolizumab monotherapy achieved PFS of 3.9 months, which is shorter than the PFS of our patients treated in the first line. ORR reported by Shoushtari, A.N. et al. is also generally numerically similar to the 23% ORR reported by D’Angelo, S.P
et al. for 86 MM patients included in another meta-analysis of multiple prospective trials of nivolumab; but seven patients—that is almost 10% of analyzed cases—were included in both described analyses, and these data may not be interpreted independently [
25,
33]. In the analysis by D’Angelo, S.P et al. covering, again, mostly patients treated with anti-PD1 ITH in further lines of therapy in NCT00730639 (CA209-003), NCT01621490 (CA209-038), NCT01721772 (CheckMate 066), NCT01721746 (CheckMate 037), and NCT01844505 (CheckMate 067) trials, ORR was—37.1% (95% CI, 21.5% to 55.1%) and the median PFS was 3.0 months (95% CI, 2.2 to 5.4 months), which represents higher ORR, but shorter PFS than reported by us [
25]. For pembrolizumab only trials—KEYNOTE-001 (NCT01295827), -002 (NCT01704287), and -006 (NCT01866319), the ORR was 22% (95% CI: 11–35) and 15% (95% CI: 5–32) in ipilimumab-naive and ipilimumab-treated MM patients, which is concordant with our treatment results. At the same time, the median PFS was 2.8 months (95% CI: 2.7–2.8) for KEYNOTE-001/-002/-006 patients, which is less than half that of the six months reported by us. Similar ITH efficacy was reported in eight patients pretreated with dacarbazine before nivolumab administration [
34]. The administration of nivolumab was listed as long-term for three patients with 13–17 cycles given, and over 30 cycles for one case. In this group, ORR was 37.5% (CR—25.0%, PR—12.5%), median PFS—10.2 months [
35].
It should be noted that in the analysis by D’Angelo, S.P et al., median PFS in patients treated with nivolumab combined with ipilimumab (
n = 35) was 5.9 months (95% CI: 2.8 months—not reached), while our patients (
n = 63) achieved median PFS of six months during anti-PD-1 monotherapy therapy. In concordance with our report, in the pooled analysis CM patients achieved PFS of 6.2 months with nivolumab monotherapy and 11.7 months with nivolumab-ipilimumab combination therapy [
25]. Nevertheless, in the pooled analysis of major clinical trials, MM patients achieved reduced clinical benefits in comparison to CM patients during anti-PD-1 ITH treatment [
25]. This was not the case in routine clinical practice reported by us with PFS in a general population of 6.9 months and OS of 20.5 and resultant HR for MM of 0.95 for PFS and 0.89 for OS. On the contrary, in clinical trial data, MM patients who were treated with nivolumab monotherapy received a median of 7.0 doses (1–34) and CM patients 11.0 doses (1–61) [
25]. Moreover, in MM patients, the median reduction in tumor burden in the target lesions was reported only at 1.4% for nivolumab monotherapy and as high as 34.2% for combined therapy [
25], which may indicate the specific biology of the disease requiring dual pathway activation for MM cell elimination.
In terms of OS of MM, patients treated with ITH were reported to achieve a median OS of 15.97 months and a one-year OS rate of 57.8% (95% CI: 49.5–67.5) while no specific data for nivolumab, pembrolizumab and ipilimumab differences were reported [
28]. In a Japanese trial, median OS was 12.0 months (95% CI: 3.5—not reached) for nivolumab treatment, while for pembrolizumab pooled analysis the median OS was 11.3 months (7.7–16.6) [
30,
34]. In general, despite the limitations of the studies discussed, data suggest that in MM patients anti-PD1 therapy may be effective both in terms of OR, PFS and OS. Nevertheless, it should be considered that in all of the multi-trial analyses, the number of MM cases was about 10% of CM patients, which may influence statistical calculations [
25]. With our report, we confirm the efficacy of ITH in MM patients with real-world data.
The efficacy of ITH may be enhanced by RT, which was reported before (ORR = 57.1%) in the case of seven patients [
36]. Another analysis of 10 patients covered mucosal melanoma of the nasal cavity or maxillary sinus, in which patients were treated with nivolumab or pembrolizumab and concomitant radiotherapy. In this group of patients, after a median follow-up period of 46 weeks, the local control rate of the primary lesion and regional lymph nodes was 100% with a median PFS of 7.4 months (range 2–82 weeks). In this group of patients, the six-month PFS rate was 60% [
37], which is lower than in our study—8.9 months. The largest report published until now was a retrospective study of 23 patients out of whom 12 patients were treated with pembrolizumab and RT, 11 patients were treated with RT alone and the others were treated with pembrolizumab monotherapy. It was shown that pembrolizumab with concomitant radiotherapy enabled achievement of a one-year target lesion control rate in 94.1% cases, while radiotherapy without immunotherapy enabled the control in 57.1% and pembrolizumab in 25%. Treatment-related AEs were not significantly different between radiotherapy with or without pembrolizumab [
38]. In our study all 23 patients received concomitant RT and immunotherapy (
Supplement C). Generally, concomitant RT is not allowed in the majority of clinical trials. Thus, a separate study with RT-ITH is required to confirm the expected benefit of irradiation during ITH. While we report the data, a separate trial is also ongoing (NCT04017897).
At the same time, no significant association is known between PFS duration and other clinical factors including primary tumor localization,
BRAF gene mutation status, stage at treatment start, presence of CNS or liver metastases, type of prior therapy, or response to first-line (ipilimumab) therapy if used in the second-line, as in the case of our analysis (
Table 2) [
31,
33]. The duration of OS and PFS are correlated with irAE development and high levels of PD-L1 expression (>5% of cells in the tumors) [
37,
38]. In fact in patients with mucosal melanoma and tumor PD-L1 expression ≥ 5% (n = 15), ORR was 53.3% (95% CI: 26.6–78.7%), while in patients with PD-L1 expression < 5% (n = 49)—ORR is 12.2% (95% CI: 4.6–24.8%), and in these patients median PFS with PD-L1 expression ≥ 5% was 12.2 months (95% CI: 3.0 months—not reached), while for PD-L1 expression < 5%—median PFS was < 3 months while the role of tumor PD-L1 expression as a response biomarker was not fully defined [
25,
39]. In fact, the significantly lower mutational burden in MM in comparison to CM may explain the decreased efficacy of the immune checkpoint blockade in many MM patients [
40,
41]. In general, not only PD-1/PD-L1 expression level, but also immune cells infiltration and transcriptional immunoscore, may be correlated with immunotherapy response. Multiple genetic and genomic factors, including mutation burden, mismatch repair deficiency, somatic copy-number variation burden and neoantigen load are being suggested to predict immunotherapy response [
42]. While somatic mutation clonality was reported to positively correlate with immunotherapy response, copy number variation (CNV) was shown to negatively correlate with efficacy of response to the PD-1 checkpoint blockade and to be associated with downregulation of immune-related pathways expression [
43,
44]. Moreover, the expression of melanin pigment and active melanogenesis, which are often found at high levels in MM, may decrease the sensitivity of MM cells to immunotherapy [
45,
46]. Melanin also has radioprotective and scavenging properties and may decrease the efficacy of radiotherapy [
47]. The relationship had not been finally defined as in survival meta-analysis pigmentation; that is, melanin level was not significantly correlated with survival (HR = 0.87; 95% CI, 0.66–1.15;
p = 0.34) in mucosal melanoma [
48].
On the other hand, it has been suggested that RT is an immune adjuvant boosting the antitumor immune response. At the same time, it is also possible that ITH has a radiosensitizing effect and increases the efficacy of radiotherapy [
49,
50]. In general, our MM study population is as large as the group that was identified in a pulled analysis of KEYNOTE-001, KEYNOTE-002 and KEYNOTE-006 trials [
50]. We confirm the effect of RTH on PFS duration in MM patients, although larger prospective trials would be needed to validate our findings. In the case of MM, only international multicenter studies would be able to recruit a large number of patients due to the epidemiology of the disease.